It is essential that the strategy for gene therapy using E. coli PNP be demonstrated in vivo. This goal will be accomplished by (1) establishing a few appropriate in vivo tumor models, (2) selecting appropriate nucleoside analogs by toxicity and pharmacokinetic evaluations, and (3) demonstrating anticancer activity of selected nucleoside analogs in the tumor models and optimizing this activity. First, the in vivo growth of tumors transduced in vitro with the PNP gene (to be supplied by Dr. Sorscher of UAB) will be characterized. The in vivo stability of the transduction will also be determined. The tumor lines that will be studied are human D-54MG glioma, rat RT-2 glioma, murine MT-539M6 glioma, murine B16 melanoma, and murine 16/C mammary adenocarcinoma. The in vivo growth of the tumors transduced in vitro with a reporter gene (to be supplied by Dr. Sorscher) will also be characterized for comparison. All growth data will be compared with that for the nontransduced parental tumor line. Second, the plasma pharmacokinetics of a nucleoside analog and its corresponding purine will be determined after a single iv or ip injection of the nucleoside. An MTD will be also determined for the nucleoside and its purine using an appropriate treatment regimen, which will depend on the plasma pharmacokinetics of the nucleoside and the in vivo stability of the transduced PNP gene. If a nucleoside analog and its purine are found to have similar MTDs, then this nucleoside/purine pair will not be used in antitumor evaluations. Third, the anticancer activity of the selected nucleoside analog (administered ip or iv) will be evaluated against a Sc implanted tumor line transduced with the PNP gene. For comparison, the evaluation will also be conducted with the tumor line transduced with a reporter gene and with the nontransduced parental tumor line. The anticancer evaluations will be repeated with l or 2 other transduced tumor line sets if available. Detailed studies to optimize and characterize the anticancer activity of the nucleoside analog (e.g., route of administration, treatment schedule, percent of transduced cells required for activity, etc.) will be conducted using the best model. The anticancer activity observed with the PNP gene therapy will be compared with that observed with standard clinical agents for the corresponding nontransduced tumor (using our historical data base).